This report provides a brief summary of the weekly ensemble forecast for Arizona from the COVID-19 Forecast Hub. In collaboration with the US CDC, our team aggregates COVID-19 forecasts from dozens of teams around the globe. Typically on Wednesday of each week, a summary of the week’s forecasts from the COVID-19 Forecast Hub appear on the official CDC COVID-19 forecasting page.
Every week, teams submit their forecasts to the COVID-19 Forecast Hub.
Each Tuesday, we combine the most recent forecasts from each team into a single “ensemble” forecast of reported COVID-19 cases at the county, state, and national level and deaths at the state and national level. At the moment, we only generate ensemble forecasts for four weeks into the future, as the available evidence suggests that models are less accurate at longer forecast horizons.
Weekly reports from the COVID-19 Forecast Hub can be found at this page.
Since the inputs to the ensemble model do not factor in changes in behavior or policy that could have an impact on short-term disease transmission (e.g. school openings or closures, new interventions, governmental policy shift, etc…), the ensemble model itself should not be looked to for specific answers to questions like “what will happen if or when schools open in 2 weeks” because most of the input models are not factoring in these changes. That said, the ensemble model has consistently shown strong predictive performance in predicting mortality, and very few of the many interventions put in place appear to have created dramatic changes in short-term disease transmission that would impact short-term trends in mortality.
This week, our ensemble combined forecasts for Arizona from 38 different models.
During periods of relative stability, models in general show broad agreement about the predicted trajectory of the outbreak. However, when there are steep surges or declines in cases or deaths, models often have quite different predictions for the upcoming weeks. The result is that there is increased uncertainty in the ensemble forecast, which can be seen as wider confidence intervals in the forecast visualization. You can explore the full set of models, including their forecasts for past weeks online at CDC’s interactive forecast visualization.
When interpreting probability of an increase, it’s important to note that the increase or decrease is relative to the location-specific recent observed incidence, which varies across locations.
At the state level, the ensemble model predicts that weekly totals of observed deaths in each of the next four weeks will be between 84 and 113 deaths each week with around 18,680 total deaths by August 28. However for the week ending August 28, the ensemble shows substantial uncertainty, with observed deaths between 40 and 270 deemed possible (95% prediction interval: 35 - 269).
For state-level death forecasts made over the last 6 weeks in Arizona for which observed data are now available, the 95% prediction intervals covered the truth 100% of the time, and the 50% prediction intervals covered the truth 100% of the time.
The sortable and searchable death table below shows the total number of reported COVID-19 deaths at the US level and state level for Arizona as of Saturday, July 31, 2021 (“Total COVID-19 Deaths”) as well as the rate of reported COVID-19 deaths in the population (standardized per 100,000 population) over the last two weeks and over the next two weeks.
The ensemble model predicts that COVID-19 daily hospitalizations will be between 139 and 260 hospitalizations daily. However for August 30, 2021, the ensemble shows substantial uncertainty, with COVID-19 daily hospitalizations between 67 and 587 deemed possible.
For state-level hospitalization forecasts made over the last 6 weeks in Arizona for which observed data are now available, the 95% prediction intervals covered the truth 66% of the time, and the 50% prediction intervals covered the truth 21% of the time. The forecasts on average over the last 6 weeks predicted values that were higher than the observed data.
The sortable and searchable hospitalization table below shows the 7 day average hospitalization count and rate of reported COVID-19 hospitalization in the population (standardized per 100,000 population) over the last week (July 25, 2021 to July 31, 2021) and two weeks ahead (August 08, 2021 to August 14, 2021).
The ensemble model predicts that weekly totals of observed cases in each of the next four weeks will be between 13,030 and 18,840 cases each week. However for the week ending August 28, the ensemble shows substantial uncertainty, with observed cases between 4,900 and 36,500 deemed possible (95% prediction interval: 4,874 - 36,474).
For state-level case forecasts made over the last 6 weeks in Arizona for which observed data are now available, the 95% prediction intervals covered the truth 61% of the time, and the 50% prediction intervals covered the truth 11% of the time. The forecasts on average over the last 6 weeks predicted values that were higher than the observed data.
Figure 2: Weekly observed and forecasted COVID-19 cases. Observed data from JHU CSSE and forecasts from the COVID-19 Forecast Hub.
The sortable and searchable case table below shows the weekly case count and rate of reported COVID-19 case in the population (standardized per 100,000 population) over the last week (July 25, 2021 to July 31, 2021) and two weeks ahead (August 08, 2021 to August 14, 2021).
County-level forecasts are only made for cases. County-level case forecasts are presented below.
Cases reported in unknown counties or by people who reside outside of the state are not included in Table 3, but are included in the state-level data.
The sortable and searchable table below shows the total number of reported COVID-19 cases at the county level for Arizona as of Saturday,July 31, 2021 (“Total COVID-19 Cases”) as well as the rate of reported COVID-19 cases in the population (standardized per 100,000 population) over the last two weeks and over the next two weeks.
For county-level case forecasts made over the last 6 weeks in Arizona for which observed data are now available, the 95% prediction intervals covered the truth 61% of the time and the 50% prediction intervals covered the truth 11% of the time. The forecasts on average over the last 6 weeks predicted values that were higher than the observed data.
Looking at the rates allows for easier comparison across counties, where you can see which counties have had or are predicted to have propoportionally higher rates in comparison to other counties. These tables calculate the weekly case count and rate of reported COVID-19 cases in the population (standardized per 100,000 population) over the last week(July 25, 2021 to July 31, 2021) and two weeks ahead (August 08, 2021 to August 14, 2021).
When interpreting probability of an increase, it’s important to note that the increase or decrease is relative to the location-specific recent observed incidence, which varies across locations.
This report was reproducibly and dynamically generated using RMarkdown. The code for the report can be found here.